The functionality of robot aids in stroke rehabilitation has been established via research involving the MIT-MANUS robot, a 2 degrees-of-freedom (dof) planar device. These studies consistently showed that robotic therapy significantly reduced impairment--by a factor of two or more by one measure--and improved recovery of function. Our company is the exclusive MIT licensee for robot therapy technology, and has developed a commercial version of the MIT-MANUS, which is now in use at hospitals and research institutions in the U S and Canada. In view of the demonstrated success of our 2-dof planar robot, a logical next step is development of a three-dimensional (3-D) upper extremity robotic therapy device. We anticipate that a spatial rehabilitation robot will accord additional impairment reduction, particularly for gravity compensation, and extend robotic therapy to encompass added functional movements with a likely greater impact on recovery of function. At present, no low-impedance robots showing positive clinical results are available for upper extremity therapy. Most researchers in rehabilitation robotics have elected to adapt or re-configure industrial robots for use in this application. While this appears reasonable, experience with industrial robots teaches us that low impedance comparable to the human arm cannot practically be achieved with these machines. An alpha-prototype of a low-impedance spatial rehabilitation robot has recently been completed. We therefore propose to refine that innovative design for commercial viability and suitability for clinical use. Toward that end, we propose to (1) Investigate inertia reduction of the alpha-prototype spatial robot, (2) Investigate the use of alternative novel linear motor technology as a direct substitute for the alpha-prototype actuator package, and (3) Redesign the alpha-prototype spatial robot. The resulting system will facilitate more efficient delivery of existing therapeutic interventions, as well as the development of enhanced and perhaps novel interventions based on the device's finely controllable interaction with patients and its data management capabilities.